Hydrogen elimination kinetics during chemical vapor deposition of silica films

Article abstract of DOI:10.1021/jp0255381

Hydrogen elimination mechanisms involved in the low-pressure chemical vapor deposition of silica films from SiH₄/O₂/N₂ mixtures are investigated. The main purpose of this work is to elucidate the mechanisms that limit hydrogen elimination in the silica deposition process under the mass-transport and gas-phase kinetic regimes. To this end, different gas-phase, surface, and mass-diffusion processes relevant to the SiH₄ oxidation and silica growth chemistry are considered and discussed on the basis of the influence of temperature, total gas flow rate and O₂-to-SiH₄ flow ratio on deposition rate and hydrogen content in the film, as determined by infrared spectroscopy and elastic recoil detection analysis. Our results indicate a clear relationship between the growth and hydrogen elimination kinetics, and support a hydrogen elimination model based on radical-surface interactions.

Full text of DOI:10.1021/jp0255381

6258
J. Phys. Chem. B 2002, 106, 6258-6264
Hydrogen Elimination Kinetics during Chemical Vapor Deposition of Silica Films
F. Ojeda,†,‡ F. Abel, and J. M. Albella*
§
,‡
Departamento de F ´ı sica e Ingenier ´ı a de Superficies, Instituto de Ciencia de Materiales de Madrid (CSIC),
Cantoblanco, 28049 Madrid, Spain, and Groupe de Physique des Solides, UniVersit e´ s Paris 7 et Paris 6,
2
Place Jussieu, 75251 Paris Cedex 05, France
ReceiVed: January 24, 2002; In Final Form: April 19, 2002
Hydrogen elimination mechanisms involved in the low-pressure chemical vapor deposition of silica films
from SiH
that limit hydrogen elimination in the silica deposition process under the mass-transport and gas-phase kinetic
regimes. To this end, different gas-phase, surface, and mass-diffusion processes relevant to the SiH oxidation
and silica growth chemistry are considered and discussed on the basis of the influence of temperature, total
gas flow rate and O -to-SiH flow ratio on deposition rate and hydrogen content in the film, as determined
4 2 2
/O /N mixtures are investigated. The main purpose of this work is to elucidate the mechanisms
4
2
4
by infrared spectroscopy and elastic recoil detection analysis. Our results indicate a clear relationship between
the growth and hydrogen elimination kinetics, and support a hydrogen elimination model based on radical-
surface interactions.
Introduction
active chain carriers (H, O, OH, and HO2). The heterogeneous
and homogeneous quenching of the chain carriers, and even of
Silicon oxides films deposited from thermally activated SiH4/
O2 mixtures are commonly used as intermetal insulator, passi-
vation layer, and diffusion barrier in the manufacture of silicon
the SiH3 species, determines the lower and upper critical ignition
limits of silane oxidation, respectively. These schemes are
consistent with the observed stoichiometry of SiH4/O2 explo-
1
-4
and II-VI/III-V semiconductor-based electronic devices.
5
6
sions, ignition limits of silane oxidation, and chromatography
measurements and photoemission spectra of silane/oxygen
The most advantageous characteristic of the SiH4/O2 chemical
vapor deposition (CVD) technique is the possibility of obtaining
reasonably high deposition rates at very low temperatures
7
flames.
Previous kinetic studies8
-15
indicate that silica growth from
(typically in the 300-450 °C range). At present, there is little
the SiH4/O2 reaction proceeds predominantly through SiOmHn
intermediate species formed in the gas phase, with negligible
contribution of surface reactions between SiH4 and O2 mol-
ecules. The low gas-phase concentration of the SiOmHn species
and their short lifetime make direct experimental observations
a very difficult task, which explains the scarce number of studies
dealing with this problem. Using in situ mass spectrometry,
question that this characteristic is due to the abundant formation
of highly reactive free radicals through a branching-chain
process in which SiH4 molecules are attacked by H, O, OH,
and HO2 radicals (active chain carriers), yielding SiH3 species
(
reactions 1-4) that ensure the chain feedback through different
oxidation channels (reactions 5-7):
1
2
Liehr and Cohen detected HnSiO species (with n g 1) in pure
SiH + H f SiH + H
2
(1)
(2)
(3)
(4)
(5)
(6)
(7)
4
3
-
3
SiH4/O2 mixtures at very low pressures (10 Torr). Van de
SiH + OH f SiH + H O
14
4
3
2
Weijer and Zwerver reported the observation of SiO radicals
in the gas phase by laser-induced fluorescence (LIF) during the
thermal deposition of silica films from pure SiH4/O2 mixtures
at 0.97 Torr. However, the chemical identity of the SiOmHn pre-
cursor species and the chemical mechanisms through which they
are produced remains unsolved to date. What is known is that
a certain amount of these hydrogenated species are trapped by
the growing oxide without being completely dehydrogenated,
leading to nonnegligible concentrations of OH and SiH groups
incorporated in the film bulk. In this sense, silica growth
involves a parallel hydrogen elimination process by which SiH4
molecules contributing to film deposition lose most of their H
atoms.
SiH + O f SiH + OH
4
3
SiH + HO f SiH + H O
2
4
2
3
2
SiH + O f HSiOOH + H
3
2
SiH + O f H SiO + OH
3
2
2
SiH + O f H SiO + O
3
2
3
This set of primary reactions branches out through the
HSiOOH, H2SiO, and H3SiO species into a series of secondary
reaction steps, not fully understood to date, yielding silica film
precursor species with the general formula SiOmHn, silica
powder precursor species, byproducts (H2, H2O, and H2O2) and
Noneliminated OH and SiH groups are one the main reasons
for the degradation and failure of SiO2-based solid-state devices.
Nonbridging SiOH and SiH groups break the network continuity,
resulting in porous structures and defects into the oxide matrix
and thus in poor electrical, barrier, protective, and optical
*
To whom correspondence should be addressed. Phone: 34 91 3349080.
Fax: 34 91 3720623. E-mail: albella@icmm.csic.es.
†
Present address: Technological & Business Development, Non-
1
2,16-23
Destructive Testing, Tecnatom S.A., Avda. Montes de Oca, 1, 28709 San
Sebasti a´ n de los Reyes, Madrid, Spain. Phone: 34 91 659 8659. E-mail:
fojeda@tecnatom.es.
properties.
In silica CVD, including SiH4/O2 CVD,
hydrogen elimination is commonly enhanced by increasing
‡
§
temperature or the oxidant-to-hydride ratio in the gas mixture
Instituto de Ciencia de Materiales de Madrid (CSIC).
Universit e´ s Paris 7 et Paris 6.
9,12,20
or by decreasing deposition rate or total pressure.
Apart
1
0.1021/jp0255381 CCC: $22.00 © 2002 American Chemical Society
Published on Web 05/22/2002

H Elimination Kinetics during CVD of Silica Films
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6259
4 2
Figure 1. Growth and hydrogen elimination mechanisms relevant to the SiH /O CVD system.
from the fact that these empirical methods do not always
provoke the desired effect, little is known about the mechanisms
causing and limiting hydrogen elimination during the CVD
process, as well as about their connection with the film growth
kinetics and the chemical paths involved in this process.
In a recent paper we have reported the influence of hydrogen
incorporation on the structure and stoichiometry of silica films
grown by SiH4/O2 CVD.24 The present work focuses on a kinetic
analysis of the mechanisms that limit hydrogen elimination
during this deposition process under mass-transport and reaction-
rate limited growth regimes. To this end, we have analyzed
jointly the influence of the experimental parameters on deposi-
tion rate and hydrogen concentration in the film. Different hy-
drogen elimination mechanisms, occurring both in the gas phase
and at the film surface, are discussed on the basis of our exper-
imental observations. This study provides, on one hand, general
methods useful to minimize and control the hydrogen content
in the film and, on the other hand, analysis criteria to study the
hydrogen elimination kinetics in other deposition processes.
transmittance spectra were recorded using a double-beam
dispersive spectrophotometer (Hitachi 270-50) in normal inci-
dence mode. The OH:SiOSi bond ratio in the film was obtained
from the ratio between the areas behind the O-H and Si-O-
-
1
Si stretching absorbance bands (found at 3760 and 1070 cm ,
respectively), according to the procedure described elsewhere.24
The stretching Si-H absorption band, typically located at 2250
-1
25
cm in silica, was always absent in our IR spectra. Therefore,
the OH:SiOSi ratio (accounting for both SiOH and H2O species)
was considered to be a good measure of the total hydrogen
2
4
concentration incorporated in the film.
Hydrogen depth profiles across the silica films were obtained
by ERDA (elastic recoil detection analysis) in the 2.5 MeV Van
2
6,27
de Graaf accelerator at the Groupe de Physique des Solides.
4
A monochromatic He beam with energies ranging from 1.9
up to 2.2 MeV was used in the usual ERDA geometry (75° tilt
of the samples). A 10 µm Mylar absorber was placed in front
of the Si surface barrier detector, located at θlab ) 30°, to stop
4
the forward-scattered He particles. Energy and composition
calibrations were performed by means of totally hydrogenated,
C8H8)n, and deuterated, (C8D8)n, polystyrene thin film refer-
ences prepared by spin-coating. These reference targets are
stable enough under our experimental conditions to be used as
standard targets for hydrogen quantitative analysis.28 The H
content of these samples was determined via their carbon content
using the procedure described in ref 29. The stability of the
silica samples was checked by means of a multichannel analyzer,
indicating hydrogen losses always better than 5%.
Experimental Section
(
Hydrogenated silica thin films were chemically vapor de-
posited on monocrystalline (100) oriented silicon wafers (Vir-
ginia Semiconductors) polished on both faces. The deposition
system was a low-pressure hot-wall horizontal CVD reactor
15
described elsewhere. This reactor was fabricated from a quartz
tube with an inner diameter of 6.1 cm and a total length of 100
cm. The length of the heating zone, which is centered in a 30
cm long cylindrical furnace, is 10 cm. A horizontal graphite
plate was used as substrate-holder. The flow of the reactants,
SiH4 (diluted at 2% in N2, 99.999% purity) and O2 (99.9992%
purity), was adjusted by means of mass flow controllers. To
prevent gas-phase reactions in the front zone of the tube, silane
was separately introduced through a stainless steel injector up
to the entrance of the reaction zone. Low-pressure conditions
were achieved by means of a vacuum system that consists of a
rotary pump boosted by a Roots pump. The base pressure, as
Hydrogen Elimination Mechanisms
To define the framework for further discussion of our results,
it is pertinent to give first a brief overview of the physicochem-
ical mechanisms relevant to the hydrogen elimination process
in the SiH4/O2 CVD system. Among them, we can distinguish
the following four (see Figure 1).
(a) Production of Precursor Species and Radicals. On the
8
-15
basis of previous studies,
the film is considered to grow
-
4
measured by a Penning gauge, was 5 × 10 Torr. The process
pressure was measured by a capacitative gauge and controlled
by means of an electronically driven throttle valve. The
deposition rate was estimated from thickness measurements
obtained by means of a profilometer (Dektak 3030).
predominantly through film precursor species (SiOmHn) formed
in the gas phase, whereas direct adsorption and subsequent
surface reactions of SiH4 contributing to film deposition are
supposed to be negligible. Film precursor species (SiOmHn) and
H, O, OH, and HO2 radicals (R) are produced simultaneously
in the gas phase by the chain reactions described above, which
can be synthesized as follows:
IRS (infrared spectroscopy) was employed to study ex situ
the hydrogen concentration incorporated in the films. IR-

6260 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Ojeda et al.
SiH + R f SiH + By
(a1)
(a2)
4
3
SiH + O f SiO H + R
3
2
m
n
where By represents reaction byproducts (i.e., H2, H2O, and
H2O2). Note that reaction a1 stands for reactions 1-4, whereas
reaction a2 includes reactions 5-7. The SiOmHn precusors
species in reaction a2 could be of the form HSiOOH, H2SiO,
or H3SiO.5
-7
(b) Gas-Phase Elimination. Self-decomposition (b1), oxida-
tion (b2), and reactions with SiH4 molecules (b3) are considered
to be the most plausible hydrogen elimination routes in the
branching stage of SiH4 oxidation:5
,13,30-33
SiO H f SiO H + By
(b1)
(b2)
(b3)
m
n
p
q
SiO H + O f SiO H + By
m
n
2
p
q
Figure 2. Arrhenius plots of deposition rate and OH:SiOSi bond ratio,
SiO H + SiH f SiO H + By
m
n
4
p
q
as determined by IRS, for silica films deposited at P ) 1.4 Torr, F )
T
2
2 4
50 sccm, and [O ]:[SiH ] ) 15.
where SiOpHq represents a less hydrogenated species with
q < n.
genating species (O , SiH , or radicals) toward the film surface
(c2-c4), diffusion of adsorbed byproducts (H O*, H *, or
2 2
H O *) toward the gas stream (c5), or surface elimination
2
4
(c) Gas-Phase Diffusion. In a hot-wall, laminar regime CVD
2
2
reactor, the mass transport in the gas phase is considered to be
controlled by the diffusion of species, driven by concentration
gradients, through a stagnant layer. Film precursor species (c1),
radicals (c2), and O2 (c3) and SiH4 (c4) molecules diffuse
through this stagnant layer from the gas stream to the film
surface, where they can further adsorb and react as described
below. In the same way, the adsorbed byproducts (By*) resulting
from these surface reactions can desorb and diffuse through the
stagnant layer toward the gas stream (c5).
reactions (within d1 or d2).
Results and Discussion
The mechanisms described above are discussed in this section
according to the growth kinetics and influence of the experi-
mental parameters (temperature, T, total gas flow rate, FT, and
oxygen-to-silane flow ratio, [O2]:[SiH4]) on the hydrogen
concentration in the film (given by the OH:SiOSi bond ratio).
The main goal is to identify the mechanisms limiting the
hydrogen elimination process operating in the SiH4/O2 system,
with the eventual purpose of concluding a simple kinetic model
for this process.
Influence of Temperature. The influence of temperature on
the deposition rate as well as on the resulting OH:SiOSi bond
ratio in the films has been studied from experiments carried
out at a pressure P ) 1.4 Torr, total gas flow rate FT ) 250
sccm, and gas flow ratio [O2]:[SiH4] ) 15. The results are shown
in Figure 2, where the deposition rate and the OH:SiOSi bond
ratio have been represented in Arrhenius-type plots. In the
deposition rate plot we can distinguish three well-defined
(d) Surface Elimination. At present, the heterogeneous
phenomena is the least understood issue in SiH4 oxidation, in
part due to the fact that most of the studies dealing with this
reaction has been devoted to its gas-phase chemistry. However,
our further discussion can be developed from two simple sets
of surface reactions: dehydrogenation of SiOm′Hn′* species
adsorbed on the film surface (d1) and of tSisOH surface
groups (d2). The first set may include self-decomposition,
oxidation, reactions with SiH4* molecules, and reactions with
H,* O*, OH*, and HO2* radicals, yielding less hydrogenated
species in the adsorbed state (SiOp′Hq′*, with q′ < n′), adsorbed
H2O, H2, and H2O2 byproducts (By*), and SiH3* radicals in
the case of reactions with SiH4* molecules:
15
regions: an activated region between 325 and 400 °C, a weaker
dependence region between 400 and 550 °C, and a decreasing
region between 550 and 800 °C. The activated region in the
low-temperature range was identified with a kinetically con-
trolled process, with an apparent activation energy of (1.41 (
SiO H * + (O *, SiH *, R*) f SiO H * + (By*, SiH *)
m′ n′
2
4
p′ q′
3
(d1)
The second set may include reactions of tSisOH groups with
incoming radicals (H*, O*, OH*, and HO2*), O2* and SiH4*
molecules, and adjacent tSi-OH groups (dehydroxylation
reactions), to give byproducts (H2O*, H2*, and H2O2*) and,
possibly, additional radicals (e.g., OH* or HO2*):
0
.2) eV. This activation energy was concluded to correspond
to some elementary gas-phase reaction, included in the global
reaction a2, where different SiOmHn intermediate species are
produced (gas-phase kinetic regime). The weak T-dependence
region was assigned to a mass-transfer regime (growth limited
by step c1). The decreasing region was associated with the
depletion of silane in the reaction zone, enhanced by the increase
of temperature and by the progressive widening of the temper-
ature profile along the reactor tube. The thermal pyrolysis of
SiH4, yielding SiH2 species, which becomes significant above
tSisOH + (R*, O *, SiH *, OHsSit) f
2
4
tSisOsSit + (By*, R*) (d2)
Those SiOH surface groups from which hydrogen is not
eliminated and those hydrogenated species (intermediate precur-
sors and byproducts) remaining adsorbed on the surface become
finally buried by the growing film and thus incorporated into
it. Thus, within the framework of this model, hydrogen
elimination can be limited by any of the following mecha-
nisms: gas-phase production of dehydrogenating radicals (a2),
gas-phase elimination reactions (b1-b3), diffusion of dehydro-
7,31,34,35
500 °C,
could enhance these depletion effects and explain
the sudden decrease of the deposition rate observed above 550
°C.
Figure 2 also shows that below ∼420 °C the OH:SiOSi bond
ratio increases abruptly with temperature, which means that
hydrogen incorporation is being controlled by an activated

mechanism strongly dependent on temperature: e.g., gas-phase
production of dehydrogenating radicals (a2), gas-phase elimina-
tion (b1-b3), or surface elimination reactions (d1 or d2). Taking
into account the weak T-dependence of the gas-phase diffusion
coefficients (typically Dg ∝ T , with R ) 1.5-2), none of the
gas-phase diffusion steps relevant to hydrogen elimination (c2-
c5) is expected to control this process below 420 °C.
Above ∼420 °C, the OH:SiOSi ratio is practically insensitive
to temperature. In this case, hydrogen elimination could be
limited either by mass transport (c2-c5 diffusion channels) or
by a nonactivated (or low activation energy) reaction. In
principle, either of the gas-phase or surface reactions described
above could fulfill this condition, i.e., a2, b1-b3, d1, or d2.
ERDA was employed to study in more detail the hydrogen
incorporation in the intermediate- and high-temperature ranges
surface and film/substrate interface) during the deposition
experiment, although incompletely for its 20 min of duration
(since the furnace was immediately removed from the reaction
zone just after each deposition experiment, the sample temper-
ature decreased to 400 °C in a few minutes and thus the time
relevant to diffusion is close to 20 min in all cases). Hydrogen
is also expected to accumulate at the SiO2/Si interface, not
observed in the ERDA spectra due to the high thickness of the
films (>8000 Å).
R
36
Hydrogen diffusion in hydroxylated silica is generally
described as the diffusion of H2O molecules through a mi-
croporous material, a complex process involving -H migration
by reactions between SiOH and SiOSi groups, reversible
dehydroxylation reactions with ionic exchange, H2O adsorption
on chemically heterogeneous surfaces, capillary condensation,
(
390 °C < T < 800 °C), since above 500 °C the hydrogen
content measured by IRS was very close to the detection limit
of the spectrophotometer, i.e., the signal-to-noise ratio for the
O-H band became very high. From the ERDA spectra we have
obtained the H/Si atomic ratio of the film. The results,
represented in Figure 3, reveal that the weak T-dependence
regime for hydrogen elimination ends up around 470 °C. Beyond
this temperature, the hydrogen concentration starts again to
decrease abruptly with temperature, suggesting the beginning
of a new regime for hydrogen elimination. This behavior may
be related to the SiH4 pyrolysis mentioned above, since the
presence of SiH2 species largely modifies the chemical path of
38-45
and bulk and surface H2O migration.
The overall diffusion
process is enhanced by temperature, which results in an
increasing hydrogen accumulation at the film surface up to 600
°C for 20 min of deposition. This effect competes with the
decrease of the film thickness due to the lower deposition rate
(see Figure 2), leading to a decreasing hydrogen accumulation
above 600 °C. Therefore, at high temperatures (T > 500 °C),
hydrogen bulk diffusion seems to influence the hydrogen
concentration profile across the film for a given experiment time,
7
,31
SiH4 oxidation.
Accordingly, SiH2 species could induce the
formation of new precursor species with a lower hydrogen
content in their formulas, and/or raise the gas-phase concentra-
tion of dehydrogenating radicals. However, the hydrogen depth
profiles obtained form the ERDA spectra (see below) revealed
a certain hydrogen accumulation at the outermost film surface
for deposition temperatures just above 470 °C, which suggests
a different explanation for this new regime.
ERDA spectra corresponding to films grown at 425, 470, 600,
and 800 °C for 20 min are represented in Figure 4. The spectrum
shape of the 425 °C film corresponds to a homogeneous depth
concentration profile. For temperatures of 470 °C and higher,
the spectra show a surface peak with increasing intensity up to
6
00 °C, and with decreasing intensity beyond this temperature.
From the simulation of the 600 °C spectrum with the senras
3
7
software, the H/Si atomic ratio was deduced to be 0.630 at
the film surface and 0.250 in the bulk. This hydrogen accumula-
tion suggests that the buried hydrogen is able to diffuse through
the film toward zones of low hydrogen concentration (film

growth kinetic regimes. The deposition experiments were
performed at temperatures of 370 and 410 °C (see Figure 5).
The [O2]:[SiH4] flow ratio was fixed at 15 and total pressure at
1.4 Torr. For both temperatures, the deposition rate shows a
maximum value for a given total gas flow rate, a behavior that
is characteristic of CVD processes governed by gas-phase
intermediate species.1
5,46,47
In the low-FT region, the deposition
rate is limited by the mass-transport rate (diffusion step c1),
which increases with FT as a consequence of the thinner stagnant
layer through which the SiOmHn intermediate species must
4
8
diffuse (the stagnant layer thickness varies approximately as
-1/2
).49
FT

H Elimination Kinetics during CVD of Silica Films
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6263
The hydrogen concentration shows again a behavior practi-
cally opposite to that of deposition rate. For [O2]:[SiH4] < 2,
the OH:SiOSi ratio decreases rapidly with increasing [O2]:[SiH4]
ratio, whereas for [O2]:[SiH4] > 2, it is nearly constant. When
the O2 molar fraction in the mixture is low, the production of
the SiOmHn intermediate species in the gas phase (reaction a2)
is also low, resulting in a low deposition rate. At the same time,
the production of dehydrogenating radicals (a2) and the SiOmHn-
into-SiOpHq conversion efficiency by gas-phase oxidation (b2)
are low. Therefore, either of both mechanisms could limit
hydrogen elimination under the gas-phase kinetic regime and
explain the high concentration of SiOH groups found in the
films for [O2]:[SiH4] < 2. By contrast, SiOmHn decomposition
reactions (b1) are not expected to depend significantly on the
O2 molar fraction, and thus can be discarded as a limiting
mechanism for hydrogen elimination under this regime. The
same argument could be applied to dehydrogenation reactions
of the type SiOmHn + SiH4 ) SiOmHn-1 + SiH3 + H2 (b3),
since the SiH4 molar fraction remains practically unchanged,
as mentioned above.
Figure 6. Influence of [O
OH:SiOSi bond ratio, as determined by IRS, under the gas-phase kinetic
regime (F ) 500 sccm, T ) 450 °C, and P ) 1.3 Torr).
2 4
]:[SiH ] flow ratio on deposition rate and
T
involved in hydrogen elimination (i.e., H2O, H2, H2O2, H, O,
OH or HO2, O2, or SiH4 species) diffuse faster than the SiOmHn
precursor species, as expected from the lower molecular masses
of the former. In any case, note that the diffusion of O2
molecules from the gas phase to the film surface (c3) can be
discarded as a limiting step for hydrogen elimination since the
high [O2]:[SiH4] flow ratio used in these experimental series
ensures an excess of O2 molecules at the film surface for the
dehydrogenation of adsorbed SiOm′Hn′ species (d1) and/or SiOH
surface groups (d2).
Influence of Oxygen-to-Silane Flow Ratio. To gain a deeper
insight into the relevant mechanisms to hydrogen elimination,
we have studied the influence of the [O2]:[SiH4] flow ratio on
the hydrogen concentration under the gas-phase kinetic regime.
In the SiH4/O2 branching-chain reactions, there is a unique
combination of O2 and SiH4 molar fractions for a given
Mixtures richer in O2 enhance the production of SiOmHn
species and dehydrogenating radicals and facilitate the SiOmHn
oxidation, being plausible explanations for the higher deposition
rate and lower incorporation of SiOH groups into the film
observed for higher [O2]:[SiH4] ratios. For [O2]:[SiH4] > 2, the
production of SiOmHn species tends to saturation. Therefore,
the gas-phase production of dehydrogenating radicals, which
is accompanied by the SiOmHn production in reaction a2, is also
expected to be quite constant for [O2]:[SiH4] > 2, being a
plausible candidate to control hydrogen elimination under the
gas-phase kinetic regime. By contrast, while the SiOmHn
production is maintained, the O2 excess should make the gas-
phase oxidation of these species more efficient. Therefore, the
fact that the OH:SiOSi ratio is not significantly reduced for [O2]:
[SiH4] > 2 allows us to reject the gas-phase oxidation of SiOmHn
temperature and pressure for which the production of intermedi-
species (reaction b2) as a limiting factor for hydrogen elimina-
tion under the gas-phase kinetic regime.
ate species is maximized.6
,31,52-54
The SiOmHn production is
reduced as the molar fraction of either O2 or SiH4 is increased
or even completely inhibited when either of them is high enough.
Figure 6 shows the variation with the [O2]:[SiH4] ratio of
deposition rate and OH:SiOSi ratio at a high enough FT (500
sccm) to allow us to operate under the gas-phase kinetic regime.
The total pressure in the reactor was maintained at 1.3 Torr,
and the temperature, at 450 °C. In this case, the SiH4 molar
fraction is practically constant as the [O2]:[SiH4] flow ratio is
varied, since O2 flow variations are compensated mostly by N2
flow variations when FT is kept at constant value. Specifically,
as the [O2]:[SiH4] flow ratio increases from 1 to 25, the O2
molar fraction (partial pressure) increases from 0.02 (25 mTorr)
to 0.33 (433 mTorr), while SiH4 molar fraction decreases from
Final Remarks. The above results support that, under the
gas-phase kinetic growth regime, hydrogen elimination is limited
by the gas-phase production of dehydrogenating radicals through
reaction a2. In coherence, we expect the dehydrogenation of
adsorbed SiOmHn species and tSi-OH groups by reactions with
radicals colliding with the film surface to be the preferential
path for hydrogen elimination in our system. Under the mass-
transport growth regime, hydrogen elimination would be then
limited by the diffusion either of these radicals toward the film
surface (c2) or of the surface byproducts toward the gas-phase
(c5).
A revealing trend common to the whole set of our results is
0
.019 (24.7 mTorr) to 0.013 (17.4 mTorr).
that hydrogen concentration decreases with increasing deposition
rate (see Figures 2, 5, and 6), which is inconsistent with
hydrogen elimination models based on pure surface phenomena
(like H2O desorption or dehydroxylation reactions). Our kinetic
model provides an evident explanation for this behavior by
assuming that the flux of dehydrogenating radicals impinging
on the growing film surface, which increases with deposition
rate, is able to overcome the burying effect of the incident
precursor species. This model is also compatible with previous
observations on the SiH4/O2 chemistry. For instance, the
heterogeneous quenching of the active radicals on the surface
exposed to them (film surface and reactor walls) is known to
be the dominant inhibition channel of SiH4 oxidation at low
For [O2]:[SiH4] < 5, the deposition rate increases rapidly with
the [O2]:[SiH4] ratio, which means that the gas-phase production
of film precursor species is determined by the O2 molar fraction
in the mixture. For [O2]:[SiH4] > 5, the increase of the
deposition rate with the [O2]:[SiH4] ratio is much slower, which
denotes that the SiH4 consumption tends to be maximized and
that there is an excess of O2 inhibiting the gas-phase reactions
rather than contributing to the SiOmHn production. If the [O2]:
[SiH4] flow ratio were further increased, we would observe a
decay of the deposition rate until reaching an upper critical limit
above which the SiH4/O2 branching-chain reactions would be
completely inhibited.3
,8,9,55,56
Analogously, a lower critical limit
31
around [O2]:[SiH4] ) 1 below which the deposition rate vanishes
can be observed in Figure 6.
pressures, which supports a hydrogen elimination model based
on radical-surface interactions. In this regard, the SiOH surface

6264 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Ojeda et al.
groups at the reactor walls have been reported to behave as
radical termination sites that affect the lower critical limit of
SiH4 oxidation, leading to hysteresis phenomena when varying
the deposition temperature in the 300-400 °C range due to
(8) Vasilyeva, L. L.; Drozdov, V. N.; Repinsky, S. M.; Svitashev, K.
K. Thin Solid Films 1978, 55, 221.
(
9) Taft, E. A. J. Electrochem. Soc. 1979, 126, 1728.
10) Vasiliev, Y. V.; Vasilieva, L. L.; Drozdow, V. N.; Shklyaev, A.
Thin Solid Films 1981, 76, 61.
11) Watanabe, K.; Komiyama, H. J. Electrochem. Soc. 1990, 137, 1222.
(12) Liehr, M.; Cohen, S. A. Appl. Phys. Lett. 1992, 60, 198.
13) Takahashi, T.; Hagiwara, K.; Egashira, Y.; Komiyama, H. J.
(
(
variations of the wall hydroxylation degree between experiment
and experiment.15 In addition, silica deposition from SiH4/O2
mixtures has been reported to occur predominantly through only
(
Electrochem. Soc. 1996, 143, 1355.
one type of intermediate species (SiOmHn with m and n having
(14) van de Weijer, P.; Zwerver, B. H. Chem. Phys. Lett. 1989, 163, 48.
(15) Ojeda, F.; Castro-Garc ´ı a, A.; G o´ mez-Aleixandre, C.; Albella, J.
M. J. Mater. Res. 1998, 13, 2308.
unique values),11 which is in agreement with a hydrogen elimi-
nation path lacking in gas-phase dehydrogenation reactions such
as b1-b3. Neutral radicals are also known to play an important
role in the growth and hydrogen elimination mechanisms
involved in other CVD systems: e.g., hydrogen abstraction
(16) Pliskin, W. A.; Lehman, H. S. J. Electrochem. Soc. 1965, 112, 1013.
(17) Lange, P.; Shnakenerg, U.; Ullerich, S.; Schliwinski, H.-J. J. Appl.
Phys. 1990, 68, 3532.
(
18) Najmi, O.; Montero, I.; Gal a´ n, L.; Perri e` re, J.; Albella, J. M. Appl.
Surf. Sci. 1993, 70/71, 217.
19) Montero, I.; Gal a´ n, L.; Najmi, O.; Albella, J. M. Phys. ReV. B 1994,
50, 4881.
(20) Han, S. M.; Aydil, E. S. Thin Solid Films 1996, 290-291, 427.
21) Goullet, A.; Charles, C.; Garc ´ı a, P.; Turban, G. J. Appl. Phys. 1993,
57,58
59,60
reactions in diamond
and silicon
CVD (with the ap-
(
plication of different activation methods such as plasmas or hot
61
filaments) and oxygen atom induced deposition of silica films
(
in plasma-enhanced CVD from tetraethyl orthosilicate/O2
mixtures. The specific surface mechanisms accounting for the
hydrogen elimination process in our system, probably included
in reactions d1 and d2, are beyond the scope of this work and
would need extensive investigations at a fundamental level.
7
4, 6876.
(22) Griscom, D. L. MRS Bull. 1987, June 16/August 15, 20.
(
23) Ferendeci, A. M. Physical Foundations of Solid State and Electron
DeVices; McGraw-Hill: New York, 1991.
24) Ojeda, F.; Montero, I.; Abel, F.; Albella, J. M. Chem. Mater. 2001,
3, 3986.
25) Adams, A. C.; Alexander, F. B.; Capio, C. D.; Smith, T. E. J.
Electrochem. Soc. 1981, 128, 1545.
26) Amsel, G.; Nadai, J. P.; d’Artemare, E.; David, D.; Girard, E.;
Moulin, J. Nucl. Instrum. Methods 1971, 92, 481.
27) Chu, W. K.; Mayer, J. W.; Nicolet, M.-A.; Buck, T. M.; Amsel,
G. Thin Solid Films 1973, 17, 1.
(
1
(
Summary and Conclusions
(
The mechanisms limiting the hydrogen elimination process
in chemical vapor deposition of silica films obtained from SiH4/
O2 mixtures have been investigated. Our results suggest that
the film grows through intermediate SiOmHn species diffusing
toward the film surface, where they adsorb and further
incorporate, yielding SiOH surface groups. Simultaneously,
dehydrogenating radicals (H, O, OH, and HO2) produced in the
gas phase by means of the same reactions diffuse to the film
surface, where they adsorb and further eliminate hydrogen from
adsorbed species (e.g., SiOmHn molecules and/or tSiOH
groups). We have found a clear correlation between the growth
and hydrogen elimination kinetics. When the film growth is
limited by the gas-phase kinetics, which is attained at low
temperatures and/or high total gas flow rates, hydrogen elimina-
tion is limited by the gas-phase production of dehydrogenating
radicals. Likewise, when the growth is limited by the mass-
transport rate, the regime that operates at intermediate/high
temperatures and/or low total gas flow rates, hydrogen elimina-
tion is limited either by the diffusion of these radicals toward
the film surface or by the diffusion of surface byproducts (H2,
H2O, or H2O2 molecules) toward the gas stream. At very high
temperatures (>470 °C), bulk diffusion of hydrogenated species
trapped by the growing film causes a certain hydrogen ac-
cumulation at the film surface and probably at the film/substrate
interface for a given experiment time.
(
(
6.
28) Abel, F.; Quillet, V.; Schott, M. Nucl. Instrum. Meth. B 1995, 105,
8
(
29) Quillet, V.; Abel, F.; Schott, M. Nucl. Instrum. Meth. B 1993, 83,
4
7.
(30) Kondo, S.; Tokuhashi, K.; Nagai, H.; Takahashi, H.; Kaise, M.;
Sugie, M.; Aoyagi, M.; Mogi, K.; Minamino, S. J. Phys. Chem. A 1997,
1
1
01, 6015.
(31) Koda, S. Prog. Energy Combust. Sci. 1992, 18, 513.
(32) Aivazyan, R. G. Kinet. Catal. 1997, 38, 174.
(
(
33) Aivazyan, R. G.; Sinel’nikova, T. A. Kinet. Catal. 1994, 35, 17.
34) Han, J. H.; Rhee, S. W.; Moon, S. H. J. Electrochem. Soc. 1996,
43, 1996.
35) Bell, T. N.; Perkins, K. A.; Perkins, P. G. J. Phys. Chem. 1984,
88, 116.
(
(36) Morosanu, C. E. Thin Films by Chemical Vapor Deposition. In
Thin Films Science and Technology, 7; Elsevier: Amsterdam, 1990.
(
37) Vizkelethy, G. Nucl. Instrum. Methods B 1990, 45, 1.
(38) Shackelford, J. F.; Masaryk, J. S. J. Non. Cryst. Solids 1976, 21, 55.
39) Rosencher, E.; Straboni, A.; Rigo, S.; Amsel, G. Appl. Phys. Lett.
979, 34, 254.
(
1
(
40) Pfeffer, R.; Ohring, M. J. Appl. Phys. 1981, 52, 777.
(41) Mikkelsen, J. C., Jr. Appl. Phys. Lett. 1984, 45 (11), 1187.
(42) Li, S. C.; Muraka, S. P. J. Appl. Phys. 1992, 72, 4214.
(43) McGinnis, P. B.; Shelby, J. E. J. Non-Cryst. Solids 1994, 179, 185.
(44) Poindexter, E. H. J. Non-Cryst. Solids 1995, 187, 257.
(45) Doremus, R. H. J. Mater. Res. 1995, 10, 2379.
(46) Kamata, K.; Maruyama, K.; Amano, S.; Fukazawa, H. J. Mater.
Sci. Lett. 1990, 9, 316.
47) Ellis, F. B.; Gordon, R. G. J. Appl. Phys. 1983, 54, 5381.
(48) Hampden-Smith, M.; Kodas, T. T. Chem. Vap. Deposition 1995,
, 8.
49) Kern, W., In Microelectronic Materials and Processes; Levy, R.
(
Acknowledgment. This work was partially supported by the
ECSC 7220-ED/082 project and by the 07M/0710/97 project
from Comunidad Aut o´ noma de Madrid (CAM). F.O. also
gratefully acknowledges the grant financed by CAM (683/96,
BOCM 22/4/96).
1
(
A., Ed.; NATO ASI Series E; Kluwer Academic Publishers: Dordrecht,
1986; Vol. 164, p 203.
(
(
(
(
(
(
(
50) Weiller, B. H. Mater. Res. Soc. Symp. Proc. 1994, 334, 379.
51) Keyser, L. F. J. Phys. Chem. 1984, 88, 4750.
52) Emel e´ us, H. J.; Stewart, K. J. Chem. Soc. (London) 1935, 1182.
53) Emel e´ us, H. J.; Stewart, K. J. Chem. Soc. (London) 1936, 677.
54) Azatyan, V. V.; Aivazyan, R. G. Kinet. Catal. 1991, 32, 1149.
55) Cobianu, C.; Pavelescu, C. J. Electrochem. Soc. 1983, 130, 1888.
56) Graham, J. High Temperatures-High Pressures 1974, 6, 577.
References and Notes
(1) Jensen, K. F.; Kern, W. In Thin Film Processes II; Vossen, J. L.,
Kern, W., Eds.; Academic Press: Boston, 1991.
(2) Lakhani, A. A. Solid-State Electronics 1984, 27, 921.
(
3) Bennett, B. R.; Lorenzo, J. P.; Vaccaro, K. Appl. Phys. Lett. 1987,
(57) Butler, J. E.; Woodin, R. L. Philos. Trans. R. Soc. London A 1993,
342, 209.
(58) Angus, J. C.; Wang, Y.; Sunkara, M. Annu. ReV. Mater. Sci. 1991,
21, 221.
(59) Kratzer, P. J. Chem. Phys. 1997, 106, 6752.
(60) Nakajima, K.; Miyazaki, K.; Koinuma, H.; Sato, K. J. Appl. Phys.
1998, 84, 606.
5
0, 197.
4) Katz, A.; Feingold, A.; Chakrabarti, U. K.; Pearton, S. J. Appl.
Phys. Lett. 1991, 59, 2552.
5) Hartman, J. R.; Famil-Ghiriha, J.; Ring, M. A.; O’Neal, H. E.
Combust. Flame 1987, 68, 43.
6) Kondo, S.; Tokuhashi, K.; Nagai, H.; Iwasaka, M.; Kaise, M.
Combust. Flame 1995, 101, 170.
7) Koda, S.; Fujiwara, O. Proc. Symp. Int. Combust., 21st 1986, 1861.
(
(
(
(61) Raupp, G. B.; Cale, T. S.; Hey H. P. W. J. Vac. Sci. Technol. B
1992, 10, 37.
(